Confident compound identifications is still one of the major bottlenecks in metabolomics. While there are ongoing efforts to refine the definitions and levels of metabolite identifications that were first proposed by the MSI initiative, it is clear that higher levels of identification confidence can be reached by joining accurate measurement technology, orthogonal molecular features, and sophisticated software tools. Here we present a single integrated software solution for pushing the confidence in identifications at different levels: molecular formula, compound class, structure, or verified targeted identification. This highly integrated functionality is implemented in a new version of the MetaboScape® software.
We could highly improve the quality of compound identifications in a study investigating the arginine biosynthesis in Corynebacterium glutamicum conducted by HRAM LC-QTOF non-targeted metabolomics. The integrated tools in MetaboScape were used to create annotations throughout increasing confidence levels: First, for all compound spectra molecular formulas were generated based on accurate masses, true isotopic patterns, and in-source fragmentation patterns, applying metabolomics-tailored rules and filters. Afterwards, public chemical databases were queried to find structural candidates for the generated molecular formulas of interesting features. Then, a customized analyte target list was applied to additionally exploit retention times of expected features. Lastly, MS/MS spectral library comparisons and in-silico fragmentations (MetFrag [1,2]) enabled to create and verify identifications based on MS/MS fragmentation patterns.
The outlined strategy will enable users to achieve highest confidence in compound identifications based on LC-HRAM-MS/MS spectral information using an integrated “turnkey” solution.